15
Rule-following, objectivity and meaning
BOB   HALE
1 Wittgenstein on meaning, understanding and rules
There is widespread agreement that Wittgenstein advances, in the rule-following sections of Philosophical Investigations and Remarks on the Foundations of Mathematics,1 considerations that are quite destructive of certain conceptions of meaning, understanding and rule-following into which we may easily slide when we attempt a general philosophical account of them: that meaning something by a certain expression is a special act or state of mind, accompanying or lying behind writing or speaking; that understanding an expression consists in supplying or adopting an interpretation for it; that following a rule - a rule for the use of a word, say - is a matter of travelling along rails which are already laid down and determine its application in new cases, and so on. And it is equally generally agreed that Wittgenstein's aims, in his discussions of these matters, are not wholly negative and destructive - that he seeks to replace these misconceptions by a better account, armed with which we shall be able to resist the pressures which push us into them: using an expression according to a rule is not founded upon reasons, but that does not mean that there can be no going right (or wrong) in our use of expressions - and the key to understanding how this can be so lies in the idea that to employ an expression with a certain meaning, or according to a rule, is to participate in a custom or practice. It is, in other words, no part of his overall purpose to uphold blanket sceptical conclusions, to the effect that there are no such things as meaning something by a particular expression, as understanding another's words or as employing an expression according to a rule (or as following a rule of any kind). His aim, rather, seems clearly enough to have been to rid us of badly mistaken pictures of what these things are, and to point us towards a proper, less inflated, conception of them.
This much is, I believe, quite uncontroversial. What is controversial is the extent of the destruction wrought by the negative considerations Wittgenstein advances and, consequentially, the exact character of the conception of meaning, understanding and rule-following - centred on the somewhat elusive ideas of custom and practice - that we may retain in the light of a proper appreciation of their destructive effect. There is, in particular, a sharp opposition between what may be termed 'conservative' readings, which see Wittgenstein as solely concerned to undermine certain seductive misconceptions2 and count it an error to interpret him as
369
BOB   HALE
providing support for any sceptical or revisionary theses about meaning and related matters, and more radical ones3 which claim to find in his writings grounds for calling into question, in one way or another, what may roughly and provisionally be called the objectivity of meaning.
This exegetical issue will not be pursued here.4 Even if the conservatives are right, the more sceptical lines of thought which Wittgenstein's discussions have suggested to some thinkers quite certainly merit careful attention in their own right, whether or not they can defensibly be attributed to Wittgenstein, or regarded as drawing out consequences of claims to which he uncontroversially commits himself. It is with two of these more sceptical directions of theorizing that we shall be concerned.
2 Kripke on rules
Kripke (1982) interprets central sections of Philosophical Investigations (§§ 138-242) as developing a 'sceptical paradox' about meaning. The paradoxical conclusion of the sceptical argument is that there is no fact about what anyone means by any expression she uses. Faced with this seemingly outrageous conclusion, we naturally incline to the view that there must be something wrong with the argument leading to it: that it relies on some assumption which we can reject, or that it makes some fallacious step. To attempt to sustain this claim is to go for a 'straight solution', which enables us to maintain that there is, after all, some species of fact in which our meaning what we do by our words consists. But Kripke argues - and takes Wittgenstein to have argued - that there can be no such meaning-constitutive facts: the argument, to be reviewed shortly, proceeds by elimination, that is, it considers the various types of fact that might be supposed to play this constitutive role, and tries to show that they cannot do the job required of them. So Kripke advocates instead a 'sceptical solution', that is, a response to the paradox which accepts the sceptical conclusion but seeks to explain how we can live with it; in particular, how we can rehabilitate talk of meaning without supposing that there are facts in virtue of which meaning ascriptions (such as statements of the form 'S means such-and-such by E') are true or false.
Kripke develops the sceptical argument in terms of one central example. Suppose '68 + 5 7 = ?' is a question I have never explicitly considered. What answer should I give? I shall almost certainly answer '125'. And I shall naturally suppose that this is not only the arithmetically correct answer, but the one I must give, if my answer is to be in accord with what I have all along meant by ' + ' or 'plus'. It is, I suppose, a fact that when I used ' +' before, I meant a certain definite function - one which has, inter alia, the value 12 5 for the arguments 68,57, and not some other function, which has a different value for those arguments. In particular it is a fact, surely, that I didn't mean the function Kripke calls 'quus' (for which we shall use the symbol '©'), where m © n = m + n, provided that m,n < 57, but in case m or n>57,m©n = 5.
Kripke's sceptic maintains that there is no such fact. His argument focuses initially on the claim that '125' is the answer I must give, if I am to be in accord with
370
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
what I formerly meant by' +'. Granting, pro tem, that there is no problem about my present understanding of' +' - that I use it to mean addition - the sceptic presses two questions, one constitutive and one epistemological:
What makes it the case that up to now I have meant addition rather than, say, quaddition by ' + ', so that '125' is the answer I should give to '68 + 57 = ?', if I am to be in agreement with what I meant by my previous uses of' + '?
What justifies me in thinking that this is the answer I ought to return, if I am to be in agreement with my past meaning for ' + '?
It is crucially important to note that these questions are posed against the background of an idealizing assumption about my cognitive powers. I am assumed to have perfect recall of all potentially relevant aspects of my past linguistic and non-linguistic behaviour, and of all my preceding mental life, any previous thoughts, imaginings, or the like, which may have accompanied my previous uses of' + '. It should be noted also that there is to be no prior restriction upon the type of fact that may be admissibly cited as constitutive of meaning; in particular, there is no Quin-ean restriction to purely physical or behavioural facts. This idealization sets Kripke's sceptic apart from the traditional variety of epistemological sceptic: if, even under the idealizing assumption, it proves impossible to justify the claim that I meant addition, the conclusion to be drawn is not - with the traditional sceptic -that whilst there may be a determinate fact about what I meant, it lies beyond our epistemic reach, but that there simply is no such fact at all.5
The sceptic's answer to both questions is, of course: 'Nothing'. By hypothesis, I have never confronted this particular addition problem before, so that the answer I should now give is not settled by my having previously had the explicit thought, or forming the explicit intention, to answer this question by '125'. Furthermore, my past applications of' +' are finite in number, and it is clearly consistent with my past answers to questions of the form 'm + n = ?' that I meant some other function by ' + ' (such as ©), which coincides with addition over the cases actually encountered, but diverges from it over '68 + 57 = ?'. No finite selection of answers determines to within uniqueness what rule (if any) I was following. The sceptic then argues that no state or event in consciousness - no previous thoughts or imaginings, nor even a special experience of meaning - can constitute the needed fact. First, it is obviously questionable whether there is in fact any single conscious state or event which invariably accompanied my previous uses. Second - and more important - even if there had been, this would be powerless to settle the question unless that state or event in consciousness were itself insusceptible of alternative, quus-like interpretations. In particular, if any past state of consciousness is to prescribe answers in particular as-yet-unencountered cases, it would have to possess a general content - a distinctive feeling or mental picture won't do, because it will never be transparent what that requires of me in new cases; rather, it would have to be something like a general thought, such as that the answer I should give to any question of the type 'm + n = ?' is the one which I obtain by counting a collection of m marbles, say, and then a disjoint collection of n marbles, and finally
371
BOB   HALE
counting the union of these two collections. But this gets us nowhere, unless we assume that there is no parallel problem about what I meant by the terms in which the general rule was formulated. We are just assuming that by 'count' I formerly meant what I now mean by that word, and did not mean quount, where quounting the union of two sets gives the same answer as counting them, provided that neither of the sub-collections has more than 56 elements; otherwise, the result of the quount is to be '5'. Keeping this perverse interpretation in play may need further perverse hypotheses about what I meant by such terms as 'union', 'subset', or 'co-extensive'. But there is no evident reason to think they can't be conjured up, with a little ingenuity.6
In effect, Kripke's point here is that sooner or later we are going to have to deal with the situation where I am supposed to have attached a certain definite meaning to certain words without giving myself an explanation of them, or rules for applying them, in general terms - so we may as well just suppose that' +' is such.
The thought is tempting that our failure to locate a meaning constituting fact in the details of my past applications of' +', or in the conscious states or events which may be supposed to have accompanied them, results from our looking in the wrong place, for a fact of the wrong sort. My meaning one thing rather than another by my words consists, it may be supposed, in my being disposed to apply them in certain ways and not in others. The attraction of this suggestion is that it can be a perfectly good fact that I was disposed to do certain things, not others, even though I did not actually do them - for the circumstances appropriate to exercise of the disposition need not have presented themselves. In particular, it could be that I was all along disposed to answer '125' to the question '57 + 68 = ?', but never actually did so, simply because no events occurred to trigger my additive disposition in this particular way. But the dispositional proposal must be rejected, Kripke argues, for two reasons. First, although it may at first appear that linguistic dispositions have the requisite generality, this is an illusion. There are potentially infinitely many questions of the form 'm + n = ?', but it just isn't true - or so Kripke claims - that for each and every one of them I was disposed to give a certain definite answer. We can only speak correctly of my being disposed to answer this way rather than that, when the numbers to be added are not too big for me to add. In this sense, our dispositions are finite. But this means that the dispositional 'solution' doesn't overcome the problem about the finiteness of actual past uses, for the class of answers I did give or would have given is still finite; and the sceptic can then undercut the proposed solution by choosing his example so that it lies beyond the reach of my additive (quadditive?) dispositions.7 Second, the dispositional proposal fails to capture the essentially normative aspect of meaning. I may well be disposed to make certain sorts of mistake when doing addition. If what I meant by ' +' is identified with what I was disposed to say, in answer to ' +' questions, then there is no room for a needed contrast between the answers I would have given and those which I should have given, the latter being those which accord with my past meaning for ' + '. Generally, the claim that some expression means such-and-such has a normative component - it is a claim about the circumstances in which it is, or would be,
372
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
correct to apply it - which evades capture by an attempted reduction of (putative) meaning-constitutive facts to dispositional facts.8
Taking it that the alternatives considered and rejected exhaust the possibilities, Kripke's sceptic concludes that there is no fact constitutive of my having meant + rather than © by ' +' in the past, and nothing that could justify my conviction that '125' is the answer I should now give to '68 + 57 = ?', if I am to be faithful to my past meaning for '+'. Furthermore, the conclusion appears to admit of straightforward generalization: if there were a fact in virtue of which I now mean + by ' + ', then - under the idealizing assumption of perfect recall, and so on -1 would be able to cite this fact to rebut tomorrow's sceptical questions. But I shall clearly be no better placed tomorrow than I am today, so there is no such fact. And clearly enough, the sceptical argument doesn't essentially concern me, or the sign ' + ', so it applies to all other language users and all other expressions. We have the sceptical paradox in full generality: 'There can be no such thing as meaning anything by any word. Each new application we make is a leap in the dark; any present intention could be interpreted so as to accord with anything we may choose to do.'9
Kripke's Wittgenstein commends a sceptical solution. The first part of the sceptical solution agrees with the sceptic that there are no facts described or misdescribed by meaning-ascriptions, but says: that doesn't matter, because such statements are not aimed at stating facts, but have a quite different, non-fact-stating role. Kripke seeks to make plausible his attribution of this idea to Wittgenstein by linking it to Wittgenstein's abandonment of the truth-conditional theory of meaning found in his Tractatus in favour of the conception of meaning as use advocated in his later writings, according to which an account of the use of a declarative sentence will comprise, in Kripke's view, a description of the conditions in which it may be appropriately asserted, together with an explanation of its role in surrounding linguistic and non-linguistic practices. The second part of the sceptical solution brings in the community. The conditions in which it is appropriate to say things like 'Jones means addition by " + "' are essentially communal - the remark is appropriately made when we have found that Jones makes statements using ' +' which are in good agreement with the things we are ourselves inclined to say. The point and role of such remarks is to acknowledge him as a fully paid-up member of the community of adders, to convey that he can be relied upon not to come up with bizarre answers (like '5') to addition problems (like '68 + 57 = ?'), and so on.
Kripke's Wittgenstein thinks community involvement is essential to provide for the normativity of meaning. If we just consider Jones on his own, all there is is his inclination to apply the word in a certain way (to respond, unhesitatingly but blindly, to addition questions with certain answers); there is nothing for his usage to be in or out of accord with. There is no room, at the level of the isolated individual's use, for the crucial distinction between what seems to him right and what is right; so that we cannot speak of right at all.1" It is only when we bring in the community, and with it the possibility of agreement and disagreement between his
373
BOB   HALE
use and that of the rest of us, that there can be a question of his applying the word rightly in a particular case. It is essential to realize that Kripke's Wittgenstein is not proposing that there is, after all, a fact constituting my meaning + by ' + ', but an essentially communal fact: if that were his position, it would clearly be vulnerable to a community-wide version of the sceptical argument, for there would then be no less a problem about what rule the community is following than there is about the individual - there is but a finite stock of previous uses of'+' by the community, and that no more determines what function was meant than does the individual's past usage, and so on.
3 Is semantic irrealism incoherent?
Kripke's argument has, quite justly, received a great deal of critical attention, mostly aimed at making out that the sceptical paradox admits of a straight solution - either one that Kripke overlooks altogether, or one that he considers but fails to rule out. These attempts may be divided into two broad groups. In the first come those which accept the assumption to which, notwithstanding his early insistence that there are to be "no limitations ... on the facts that may be cited to answer the sceptic",11 Kripke himself appears to subscribe, that putatively meaning-constitutive facts must be specifiable in non-semantic, non-intentional terms. The main contenders here - aimed at a naturalistic solution - have been attempts to uphold some more or less sophisticated version of dispositional theory, or to show that a broadly causal account of meaning and/or reference escapes the sceptical argument. It has also been claimed that even if Kripke's objections are effective against a dispositional account, they do not dispose of the view that an expression's having a certain meaning consists in its being associated with an appropriate capacity.12 Others - the second group - take issue with what they see as a substantial reductionist assumption underpinning the sceptical argument, and have accordingly sought to defend the view that semantic facts, or closely related facts about intentions, need not be reducible to facts of some other, naturalistic kind.13
Kripke himself describes the sceptical conclusion as 'insane and intolerable'.14 But he believes that it is none the less a conclusion we have to accept. The sceptical solution, he hopes, enables us to do so. Others have taken a less optimistic view of the sceptical solution, arguing that the sceptical conclusion not only appears to be but really is intolerable. If they are right then there must be something wrong with the argument to it: a straight solution of some sort must be possible. Space does not permit detailed evaluation of the various alternatives which have been canvassed here. In this section I shall, instead, examine some arguments designed to establish the incoherence of the position to which Kripke is led by the sceptical argument. First, however, it will be useful to make some remarks about what that position -semantic irrealism, as I shall call it - involves.
Meaning-statements are made by means of declarative sentences. As such, they may be asserted on their own, and they may equally figure as components in conditionals, disjunctions and other compounds. This is, arguably, by itself enough to ensure that they may with equal propriety be embedded in such contexts as 'It is
374
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
true that..." and 'That... is a fact'. It might perhaps be insisted that such embed-dings are acceptable only if the embedded sentences express claims which are subject to standards of correctness. I am not myself convinced that that is so, but even if it is, a proponent of the sceptical solution could hardly object on that score to our saying, for example, 'That Jones means addition by "plus" is a fact'. The aim of the sceptical solution is to rehabilitate (talk of) meaning in the face of the sceptical conclusion, by explaining how we can properly and correctly assert things like 'Jones means addition by "plus"'. There is no unavoidable error in ordinary talk of this kind - Kripke is not advocating an 'error theory' of meaning discourse, analogous to John Mackie's error theory of ethical discourse (see Chapter 12, realism and rrs oppositions, section 4): the error lies, rather, in prevalent philosophical (mis)interpretations, which construe meaning-statements as genuinely fact-stating or descriptive, having genuine truth-conditions.
It is, then, an obvious thought that for this very reason, it cannot be right simply to deny without qualification that meaning-statements are ever true, or that they state facts. Kripke anticipates such an objection to the sceptical solution's endorsement of the sceptical conclusion, and suggests that it may be defused by appeal to the 'redundancy' theory of truth.15 His thought seems to be that, since 'it is true that S means that p' has the same content as 'S means that p', we are doing no more in asserting the former than we are in asserting the latter, and so are saying nothing from which a proponent of the sceptical solution need dissent. But this is puzzling.16 In fact, the point seems to tell in precisely the opposite direction: just because, given a redundancy or deflationary conception of truth (and facts), 'It is true (is a fact) that S means that p' says no more than 'S means that p', there can be nothing wrong with the former - and accordingly, if the sceptical denial that meaning-statements are true or state facts is understood as involving this minimal notion of truth or fact, it must be wrong. The moral - apparently not clearly appreciated by Kripke - is that the sceptical conclusion, if it is to have even a chance of being acceptable, must be understood as invoking some more substantial conception of truth and facts. And if the sceptical solution is to have point, clarification of the more substantial notion(s) of truth and fact whose application to meaning-statements is to be denied becomes a matter of some urgency.17 Whether a telling objection to Kripke can be erected around this point will be considered later. Meanwhile, I shall reserve the term 'true' for whatever more substantial notion might be taken to be in play, and employ 'correct' for the minimal sense. The sceptical conclusion can then be understood as claiming that meaning-statements are never true, but are (at best) correct.
More than one thinker has remarked upon the close similarity between the sceptical solution's combination of meaning irrealism with an attempt to rehabilitate meaning discourse by construing it non-descriptively, and more familiar pro-jectivist attempts to save our thought and talk in other areas, such as morality, aesthetics and modality, where the apparent absence of a suitable range of truth-conferring facts seems to preclude a fully realist construal of the discourse.18 And in one way, given the essentially normative character of the notion of meaning - on which all parties are agreed - together with the plausible claim that there can be no
375
BOB   HALE
successful reduction of the normative to the purely factual, it may seem that meaning discourse is ripe for projectivist reconstruction (see Chapter 12, realism and its oppositions, section 4). On the other hand, just because a projectivist treatment of some given region of discourse is a thesis about the kind of meaning attaching to statements belonging to it, it may be doubted whether a non-factualist or projectivist approach can coherently be applied to meaning itself. It may well seem that the philosophical point and advantage of, say, a projective treatment of ethical discourse (perhaps based upon the kind of expressivist reconstrual proposed by the emotive theory) would be substantially compromised, if coupled with the thesis that meaning discourse quite generally, and so any claim about the sort of meaning possessed by ethical statements in particular, is itself not genuinely factual but projective of, say, some attitude we have. These are, of course, no more than vague misgivings. Can they be transformed into a sharp and telling objection to the sceptical solution?
John McDowell wrote:
It is natural to suppose that if one says 'There is no fact that could consititute its being the case that P', one precludes oneself from affirming that P;. .. Given this supposition, the concession that Kripke says Wittgenstein makes to the sceptic becomes a denial that I understand the 'plus' sign to mean one thing rather than another. And now - generalizing the denial - we do seem to have fallen into an abyss: 'the incredible and self-defeating conclusion, that all language is meaningless' (Kripke, 1982, p. 71). It is quite obscure how we could hope to claw ourselves back by manipulating the notion of accredited membership in a linguistic community.19
The pessimistic conclusion is, however, too swiftly drawn. As we have seen, it is a condition of the coherence of Kripke's sceptic's argument that he is working with a substantial notion of fact, one for which the correctness of 'It is a fact that P' is precisely not guaranteed merely by the assertibility of 'P': that is, a more than merely deflationary or minimal notion of fact. But for this notion we can hardly expect there to be a generally unproblematic transition from 'It is not a fact that P' to denying that P. Kripke will want to hold, on the contrary, that there will be cases in which we can correctly assert that P, when it is not a fact that P. McDowell's 'natural supposition' just begs the question against him. It may be that irrealism about meaning, in contrast with irrealist theses in other areas, such as morals or mathematics, will turn out to suffer from some distinctive species of instability. But if so, further argument is needed to disclose it. Important arguments to the purpose have been advanced by Wright and Boghossian.2"
Wright argues in two stages: (1) irrealism about meaning leads to global irrealism and (2) global irrealism is incoherent or otherwise directly objectionable. Wright's globalizing argument pivots on what he calls the meaning-truth platitude, that 'the truth value of a statement depends only upon its meaning and the state of the world in relevant respects.' In its original version, from which this formulation of the platitude is taken, it runs thus:
376
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
If the truth value of S is determined by its meaning and the state of the world in relevant respects, then non-factuality in one of the determinants can be expected to induce non-factuality in the outcome. (A rough parallel: If among the determinants of whether it is worth while going to see a certain exhibition is how well presented the leading exhibits are, then, if questions of good presentation are not considered to be entirely factual, neither is the matter of whether it is worth while going to see the exhibition.) A projectivist view of meaning is thus, it appears, going to enjoin a projectivist view of what it is for a statement to be true. Whence, unless it is, mysteriously, possible for a projective statement to sustain a biconditional with a genuinely factual statement, the disquotational schema '| P | is true if and only if P' will churn out the result that all statements are projective.21
Against global projectivism Wright advances several related but distinguishable considerations. An obvious worry, hinted at previously, is that a projectivist treatment of any particular class of statements has point only in so far as it draws a significant contrast between members of that class and other statements which are to be viewed as genuinely fact-stating, or apt for substantial truth. Relatedly, whilst a perfectly good distinction may turn out to be empty on one side, it may be doubted whether that could be an a priori matter, as would be the case with the needed distinction between fact-stating and non-fact-stating discourse if the globalizing argument is sound. Thirdly, supposing the distinction satisfactorily drawn, the projectivist will surely want to regard it as a discovery that statements in the target class are non-factual - in particular, shouldn't the statement of the conclusion of the sceptical argument be itself genuinely factual? Fourth: there will be no truths about the (Kripkean) assertion conditions of any sentences, with the result that the premisses of Kripke's version of the argument against private language (relating to the communally-oriented character of the assertion conditions of meaning-statements), and hence also its conclusion, will enjoy a merely projective character.22
Boghossian agrees with Wright that irrealism about meaning inflates into global irrealism (though he gives a somewhat different argument for this claim), but he is not persuaded that this is intrinsically objectionable; instead, he argues that meaning irrealism leads directly to self-contradiction, independently of its implicitly global character. The argument23 starts from a generalization of the point made above: that since any significant, declarative sentence is apt for truth in a merely deflationary sense, a non-factualist thesis about any class of statements cannot be understood as denying that any of those statements are true in that sense, but must be taken to involve a richer, more substantial notion of truth. The non-factualist is, as he puts it, 'committed to holding that the predicate "true" stands for some sort of language-independent property, eligibility for which will not be certified purely by the fact that a sentence is declarative and significant'.24 He then claims that a judgement that some sentence is or is not (substantially) true cannot but be a genuinely factual judgement. That is: the judgement that S is true (and likewise the judgement that S is not true) must itself be true or false, as opposed to being merely correct or incorrect. But the meaning non-factualisťs distinctive thesis is that
377
BOB   HALE
judgements about what a sentence means are not factual. Since what truth-condition a sentence possesses is a function of its meaning, it follows that judgements about what truth-condition a sentence has are likewise not factual. And since a sentence's having a particular truth-value cannot be a factual matter if its having a certain truth-condition is not, it further follows that a judgement about a sentence's truth-value can never be factual. Thus the meaning non-factualist is committed to denying that it is ever true that S is true. His position is thus self-contradictory.
Does either of these arguments succeed? Obviously enough, the crucial claim in Boghossian's argument is that the judgement that a sentence is (substantially) true must itself be a genuinely factual. But it is anything but obvious that the meaning non-factualist must agree. He is, as we have seen, committed to the intelligibility of a thick (that is, more than merely deflationary) notion of truth; though whether he is further committed to its having a non-empty extension must, at this stage, be regarded as an open question. But acceptance ofthat much seems perfectly consistent with retention of the thin, merely deflationary notion which we are calling correctness. And so long as both notions are available, why can't the non-factualist hold, apparently with perfect consistency, that metalinguistic attributions of truth, falsity, correctness and incorrectness are all alike, at most correct and never true? Boghossian believes that the non-factualist has not merely to make room for a thick (or as he says 'robust') notion of truth, but that he must choose between that and a purely deflationary one:
It is an assumption of the present paper that the concept of truth is univocal.. . We should not confuse the fact that it is now an open question whether truth is robust or deflationary for the claim that it can be both. There is no discernible plausibility in the suggestion that the concept of a correspondence between language and world and the concept of a language-bound operator of semantic ascent might both be versions of the same idea.25
Clearly so crucial an assumption stands very much in need of supporting argument; surprisingly, Boghossian provides none, unless you think that the last sentence quoted does more than merely reassert what needs to be established.
This objection coincides pretty well, I think, with one of several developed in more detail by Wright,26 who insists, as I have done, that the non-factualist is free to wield notions both of truth and correctness. Somewhat ironically, this distinction appears at first to provide the non-factualist with a ready way to interrupt Wright's own attempted reductio at the first, globalizing stage. As Wright's formulation makes plain, the final step involves an application of the Disquotation Scheme for 'true'. More specifically, he appears to have envisaged substituting the right- for the left-hand side of the scheme, to get from:
' "P" is true' is not true
to
378
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
'P' is not true
But it now appears that the non-factualist may block this step: when 'P' is true,' "P" is true' will be merely correct, so that the Disquotation Scheme - 'P' is true if and only if P - fails right-to-left.
In fact, this claim relies upon a questionable assumption about the evaluation of conditionals, that is, that a conditional will hold (be true, or at least correct) only if there is no descent in value (from true to correct, say) between its antecedent and consequent. As against this, it may plausibly be claimed that we should require only preservation of designated value (where true and correct are designated, the remaining values not). However, whilst this makes it at least doubtful that the non-factualist can block the globalizing argument by rejecting the Disquotation Scheme outright, it leaves him with the resources for an equally effective rejoinder - indeed, a more satisfactory one, because it allows him to retain the Disquotation Scheme. For if the scheme is secured by adoption of the proposal that what is required for a conditional to hold is not that the consequent is true if the antecedent is, but only that designated values shall be preserved, then instances of the biconditional scheme will not support substitution of their components in complex contexts such as that involved in the globalizing argument.27
There is another, more obvious, ground for dissatisfaction with the globalizing argument, at least in Wright's version. For it seems clear that, at least as formulated, the argument given works at best for a sense of 'statement' in which statements can be taken as both bearers of meaning and bearers of truth value. A proponent of semantic irrealism need not deny that there is, or could be, such a sense of 'statement', provided that he is granted a different sense in which statements have truth values, but cannot sensibly be said to have meanings. Concerning any statement in this sense, he can claim that whether or not it is true is a factual matter; or more precisely, that its being so is not threatened by the non-factuality of meaning. It is true enough that whether or not a particular sentence is suitable for making a particular statement in this sense depends upon the sentence's having a certain meaning - and that, he holds, is not a factual matter. But the truth value of a statement, in his preferred sense, does not depend upon the meaning of anything. It does not depend upon the meaning of the statement, because statements are not the sort of thing to have meanings; and it does not depend upon the meaning of a certain sentence - what depends upon a sentence's meaning being, rather, what statement(s) that sentence can be used to make. Thus non-factuality at the level of meaning does not induce non-factuality at the level of truth-value of statements.28
This discussion has inevitably been somewhat inconclusive. If what I have argued is right, it has not been shown that irrealism about meaning leads directly to contradiction, independently of its putative tendency to inflate into global irrealism; and it is at least open to question that it does globalize. And even if it does globalize in the way Wright and Boghossian both believe, it remains to be seen whether that leads to its collapse. Wright's arguments are suggestive of instability here, but appear less than decisive.
379
BOB   HALE
4 Wright on the rule-following considerations
4.1 The contractual model of meaning and investigation-independence
While Wright is sharply opposed to the semantic-irrealist conclusion which Kripke extracts from the Rule-following Considerations, he advances29 an argument (for which he claims Wittgensteinian origins) whose conclusion has - rightly or wrongly - been seen as carrying implications for the notion of objectivity which are scarcely less radical, and no more palatable, than Kripke's. The argument is directed not at calling in question the very existence of facts about meaning, but at undermining what Wright takes to be an important misconception of their character. According to the conception under attack - the contractual model of meaning30 - an expression's having a certain settled meaning consists in its being associated with a definite pattern of application which, once established, extends 'of itself to new cases quite without any further assistance from us. Learning what the expression means is a matter of 'cottoning on' to such a pattern; our subsequent employment of the expression then either conforms, or fails to conform, with requirements already laid down, as it were, in the contract to which we have become party. Wright's contention is that, for reasons implicit in Wittgenstein's discussions of rule-following, the contractual model is fundamentally flawed and must be replaced by a conception of meaning as shaped by our ongoing use.
That is the immediate conclusion of Wright's argument. If for no other reason, the argument which purports to establish it deserves the closest scrutiny simply because the contractual picture is one which we may find both appealing and entirely natural, and which may, indeed, seem inevitable when we seek to understand what is involved in the normativity of meaning, so that Wright's conclusion is at the very least unsettling. But Wright draws a further conclusion which may appear not merely unsettling, but plainly intolerable. This concerns the way or sense in which ordinary factual statements may be held to be objectively true or false. What, in very general terms, we intend when we take a statement to be objectively true or false, is that its truth-value is in some way independent of our, or anyone else's, opinion. But this somewhat vague idea can be cashed out in various more specific ways. We ought not to be surprised if it should prove that what more precise characterization of it is found acceptable depends upon where one's sympathies lie in the dispute between realists and anti-realists in the theory of meaning (see Chapter 12, realism and its oppositions, sections 1 and 2). Wright focuses upon one particular conception of objectivity - investigation-independence - which we might expect anyone of a realist persuasion to endorse. A realist, in Dummett's sense, about a certain class of statements - that is, one who holds that statements in that class are such that their truth-conditions may be fulfilled, or not, without our being even in principle capable of recognizing as much - evidently regards those statements as objectively true or false. But if a capacity for evidence-transcendent truth is taken as the criterion, the resultant sense of objectivity is very strong indeed. By their very nature, no effectively decidable statements will qualify
380
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
as having objective truth-values in this sense. And the same, it is natural to suppose, goes for very many other perfectly ordinary statements which, though not effectively decidable in any strict sense, would normally be viewed as capable of objective truth. Assuming that the realist wishes to regard statements of these latter kinds as capable of objective truth, what alternative criterion should he adopt? Wright's plausible suggestion31 is that he will embrace a notion according to which 'confronted with any decidable, objective issue, there is already an answer which, if we investigate the matter fully and correctly, we will arrive ať. For such statements, that is, objectivity of truth-value consists in the possession of a determinate, investigation-independent truth-value. But investigation-independence, Wright argues, requires the contractual model of meaning:
Investigation-independence requires a certain stability in our understanding of our concepts. To think, for example, of the shape of some particular unobserved object as determinate, irrespective of whether or not we ever inspect it, is to accept that there are facts about how we will, or would, assess its shape if we do, or did, so correctly, in accordance with the meaning of the expressions in our vocabulary of shapes; the putative investigation-independent fact about the object's shape is a fact about how we would describe it if on the relevant occasion we continued to use germane expressions in what we regard as the correct way ... The idea of investigation-independence thus leads us to look upon grasp of the meaning of an expression as grasp of a general pattern of use, conformity to which requires certain determinate uses in so far unconsidered cases. The pattern is thus to be thought of as extending of itseif to cases which we have yet to confront.'2
If this is correct, a successful argument against the contractual model will be equally destructive of the idea that statements are capable of objective truth-value in the sense captured by investigation-independence.
It is obvious that any statement which is evidence-transcendently true or false will be objectively true or false in the sense captured by investigation-independence, but that the converse does not hold. In that sense, the latter is a weaker notion of objectivity than the former. And this, coupled with the fact that the notion of evidence-transcendent truth plays no part in the characterization of the weaker notion, might suggest that someone who rejects realism in Dummett's sense could endorse the claim that there are investigation-independent truths. But if Wright's argument is sound, this is an illusion. For the argument, as we shall see, makes essential use of an anti-realist premiss, so that if he is unable to find fault with it elsewhere, the anti-realist must reject the notion of investigation-independent truth.
4.2 The 1980/1 argument
Can an individual speaker S, in her use of an expression E, defensibly be regarded as attempting to conform to a pattern of application, the requirements of which are already in place? Wright's argument, in its earlier version, divides the question into two.
381
BOB   HALE
First, can we defensibly regard S as aiming at conformity to such a pattern independently of the possibility of assessment of her performance by others?33 The difficulty here is to see how it can be justified to describe the situation in terms of S's recognizing what her supposed pattern requires her to say, in any particular case, as opposed to her merely being disposed to apply E (or not, as may be). The former description is justified only if there is a distinction to be drawn between S's going on as the pattern demands on the one hand, and on the other her merely seeming to do so. But S cannot make this distinction for herself, since it is bound to seem to her that her sincere and considered application of E conforms to the requirements of the pattern; and by hypothesis, the distinction is not to be made out on the basis of others' assessment of her performance.
Since the contractual picture cannot be sustained for this case, we move to the question of whether it can make a difference to the situation if we add in, as it were, facts about the agreement, or lack of it, between S and the rest of us over the application of E. Wright argues that it makes no essential difference. Here it is crucial to remember that the question at issue is not whether agreement with the community somehow provides the standard of correctness, but whether bringing in agreement, or lack of agreement, with the community affords a way of keeping the contractual picture in play. As Wright puts it, it is the question: "How does others' agreement with me turn my descriptive disposition into a matter of recognition of conformity with a pattern, recognition of an antecedent fact about how the communal pattern extends to the new case?" The answer, unstated but clearly implied, is that it cannot do so.
Wright restates this last part of the argument in a somewhat different way, which is worth noticing because it corresponds rather more closely to his later, and much terser, formulation.34 If S's agreement with the rest of us somehow made the crucial difference, so that she could be thought of as recognizing what the shared pattern dictates in a given case, then it should at least make sense for her to claim, should she find herself at loggerheads with the rest of us over the application of E, to recognize that we have gone off track. But the only proper conclusion for S to draw, given that she can find no way to persuade us that we have broken faith with our antecedent pattern, is - or so Wright contends - that she does not (and perhaps never did) know what E means (as we employ it). But if no one can recognize that the community has gone off the rails, no one can recognize that the community has stayed on them; mere lack of disgreement with the community cannot substantiate the claim to recognize what its supposed pattern requires.
It is tempting, as Wright notes, to think that "a solicitable community of assent just does make the relevant difference". But he gives a supplementary argument35 which, if good, shows that the temptation must be resisted. On the contractual model, the bearing of communal agreement over the application of E on the correctness or otherwise of S's use has to be understood in a quite particular way. It is not that communal agreement is constitutive of correct use. Correctness must consist in conformity with the requirements of the community's pattern, and communal agreement can be at best36 good inductive evidence for that. In other words, on the contractual model, a community of assent on what should be said in a given case
382
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
provides the standard against which individual applications of E are to be assessed only because and in so far as communal agreement can be taken to be based upon recognition of what the community's shared pattern requires. But once this is seen, it should be clear that we have no progress: so far from answering the objection previously urged against the picture of individual, community-independent conformity to an antecedent pattern, bringing in the community merely shifts the target. For what now requires justification is description of the situation in terms of the community's recognizing what its supposed pattern requires, in any particular case, rather than in terms of its merely being disposed, collectively and non-collu-sively, to apply E (or not, as may be). The former description is justified only if there is a distinction to be drawn between the community's going on as its pattern demands on the one hand, and on the other, its merely seeming to it that it is doing so. As Wright puts it:
If 'correctness' means ratification-independent conformity with an antecedent pattern, there is apparent absolutely nothing which we can do to make the contrast active between the consensus description and the correct description17
Of course - and as Wright agrees - we may as a community retrospectively judge that our erstwhile, communally agreed verdict on a particular case was mistaken; but this can give no comfort to the contractualist, since it is obviously wholly tendentious to view this as a matter of our belatedly recognizing that we previously broke faith with the requirements of an antecedently determinate pattern.i8 The necessary contrast between recognizing what our pattern required, and our earlier, collective disposition concerning what to say merely changing, is evidently no less problematic than that between recognizing what our pattern requires us to say now and our present disposition.
Although it will scarcely have escaped the notice of readers already familiar with this debate, it is worth underlining the argument's reliance, in its closing step at least, if not earlier, upon an anti-realist premiss to the effect that there is no sense to the claim that we operate with a distinction - in this case between the supposedly ratification-independent requirements of our pattern of use and how we think we should apply the expression in question - if there is nothing we can do to manifest a grasp of it. Wright himself is under no illusions, of course, about the need for such a premiss, and indeed, stresses the point:
If those arguments [i.e. the general anti-realist arguments against the intelligibility of attributing grasp of concepts of which there is no distinctive manifestation] are rejected, then there is... no obstacle to embracing the investigation-independence of decidable statements. If, and only if, one admits the need to describe how an understanding could be revealed of what it is for our consensus verdict... to fit the alleged investigation-independent fact of the matter ... will one feel pressured to reject the 'double-element' conception.59
That concludes my summary of Wright's argument in its earlier formulation. It leaves us facing three main questions: (1) is the argument sound? (2) is the
383
BOB   HALE
rejection of investigation-independence it enjoins tolerable? and (3) if the contractual model is to be scrapped, what should replace it?
4.3 Horrified reactions
In view of the apparent innocence of the notion of investigation-independence, it is no surprise that others have seen its rejection not as a salutory corollary of the argument against the contractual model, but as revealing that something must have gone badly wrong, either in the argument Wright builds upon Wittgenstein's discussions of rule-following or in the RFC themselves. Thus John McDowell writes:
If Wittgenstein's conclusion, as Wright interprets it, is allowed to stand, the most striking casualty is a familiar notion of objectivity. The idea at risk is the idea of things being thus and so anyway, whether or not we choose to investigate the matter in question, and whatever the outcome of any such investigation. That idea requires the conception of how things could correctly be said to be anyway - whatever, if anything, we go on to say about the matter; and this notion of correctness can only be the notion of how the pattern of application that we grasp, when we come to understand the concept in question, extends, independently of the actual outcome of any investigation, to the relevant case. So if the notion of independent-investigation is to be discarded, then so is the idea that things are, at least sometimes, thus and so anyway, independently of our ratifying the judgement that that is how they are. It seems fair to describe this extremely radical consequence as a kind of idealism.40
Although McDowell thinks that we cannot accept Wright's conclusion, and is thus committed to denying the soundness of the argument leading to it, he does not dispute its validity. In fact, he is committed to its validity, because he takes it to form the core of an effective 'transcendental argument against anti-realism' which reduces to absurdity the anti-realist premiss upon which, as we have noted, the argument relies. It would, of course, be wholly tendentious, in the present context, to rest such a reductio on the alleged absurdity of the denial of investigation-independence itself. The absurdity lies, rather - or so McDowell contends - in the picture of language to which Wright's argument commits him, on which there is no room for normativity, and so no room for meaning, at all. Of course, Wright does not himself think that his argument leads to this absurd conclusion. But there is no escaping it, McDowell claims, once we accept with Wright that at the individual level there is no going right or wrong in our use of words save in the context provided by communal assessment of individual use, and that as far as the community as a whole is concerned there is no authority to which its collectively agreed use is answerable, and no distinction to be drawn between the 'consensus description' and the 'correct description', so that we cannot say that it 'goes right or wrong', only that it 'just goes'. For this entails a picture of language use on which, 'at the basic level', human beings are merely 'vocalizing in certain ways in response to objects', no doubt to the accompaniment of certain 'feelings of constraint, or convictions of the Tightness of what they are saying', but at which 'there is no question of shared commitments - of the behaviour . . . being subject to the
384
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
authority of anything outside themselves . . . How, then, can we be entitled to view the behaviour as involving, say, calling things "yellow", rather than a mere brute meaningless sounding off?' And once we are committed to this picture of the 'basic' level, stripped of normativity altogether, there is no hope of reinstating it via the notion that individuals are subject to communal correction. As McDowell puts it,
The problem for Wright is to distinguish the position he attributes to Wittgenstein from one according to which the possibility of going out of step with our fellows gives us the illusion of being subject to norms, and consequently the illusion of entertaining and expressing meanings.41
This attempt to turn Wright's argument on its head is, it seems to me, a complete failure. McDowell plainly takes it that when Wright observes that there is no standard against which the whole community's practice may be assessed, he is advancing this as his own view. But Wright is doing no such thing; he is himself offering a reductio of the idea that correct use is a matter of conformity with a ratification-independent pattern. McDowell appears entirely to have overlooked the crucial point that the conclusion that there is no distinction between the consensus verdict and the correct verdict is drawn on that hypothesis - hence Wright's conditional: 'If "correctness" here means ratification-independent conformity with an antecedent pattern, there is apparent absolutely nothing we can do to make active the contrast between the consensus description and the correct description'. Wright's argument as I understand it is that if communal correctness were a matter of conformity with such a pattern then, unless whether or not the community goes right is to be a verification-transcendent matter, there would have to be a distinction between the community's recognizing what its pattern requires and its merely thinking that it does. Since no content can be assigned to this contrast, there can be no content either to the distinction between the consensus verdict and the correct verdict, on this supposition about what correctness consists in. Given the obvious unacceptability of the conclusion to which it leads, we should reject that supposition.
Somewhat differently, Michael Dummett, in effect,42 agrees with Wright that "an unflinching application of Wittgenstein's ideas about rules" leads us to deny that there can be pre-determinate, investigation-independent facts; since he finds this conclusion incredible, he concludes that 'the "rule-following considerations" embody a huge mistake'. Wittgenstein was
right to observe that, for the most fundamental of the rules that we follow, there is nothing by which we judge something to be a correct application of them. It certainly does not follow from this that, if we never do make such a judgement in some particular instance, there is no specific thing that would have been a correct application: to draw that inference, you need a general internalist premiss, that there is nothing to truth beyond our acknowledgement of truth.4'
But this premiss, Dummett complains, is totally implausible; to appeal to it in this context is simply to beg the question.
385
BOB   HALE
Since Dummett is discussing Wittgenstein, and not Wright, it would be unjust to complain of a failure to engage the latter's argument. But it is pertinent to observe that Wright's argument is, on the face of it, an argument of precisely the kind whose possibility Dummett denies, since it manifestly does not appeal to any premiss to the effect that there is no more to truth than its being acknowledged.
The immediate reason why Dummett finds the rejection of investigation-independence incredible is that it appears to involve denying, in the case of an elementary calculation, that there is in advance of its being carried out any deter-minately correct result. Another reason that might be given (to which Dummett attaches great importance, both in the paper from which I have quoted and in earlier writings) is that it appears impossible to account satisfactorily for the value or usefulness of deductive inference without appealing to a distinction between a statement's being true and its being actually verified or recognized as true, and hence, it may seem, without invoking the possibility of investigation-independent truth. Indeed, this may be seen as a special case of a quite general difficulty. For it may seem that if Wright's conclusion stands, nothing approaching justice can be done to the conception we all have of human enquiry in general as a process of discovery. These are matters for genuine concern, to which - so far as I have been able to see - nothing in Wright's earlier presentations of the argument speaks. Pending explanation of how it might be alleviated, we have a strong motive for hoping, if not for suspecting, that there is after all a flaw in his argument.
4.4 Wright's strengthened argument
In fact, there is a flaw in it. The first part of the argument, aimed at showing that there can be no substance to the idea that an individual speaker is aiming at conformity to a pattern of use independently of the possibility of assessment of her performance by others, seems to me compelling. We should also agree that the contractual conception requires the possibility of community-wide (and so of near-community-wide) departure from its pattern for a given expression. But Wright's next claim - that the only conclusion a lone dissenter could properly draw, on finding herself unable to bring the rest of us round, is that she no longer understands the crucial expression, and so cannot be a competent critic of the rest of the community's use - is far from clearly correct. On the contrary, it appears that there is plenty of room for a proponent of the contractual view to resist it. No doubt I -should be disconcerted to find the rest of the community lined up against me. But it is far from self-evident that, were this to happen, it must be that í have gone astray, and cannot be that they have done so. We can surely envisage circumstances in which the opposite would be the case. It is, for instance, at least conceivable that everyone else has, perhaps as a result of exposure to some insidious form of radiation which I have escaped, suffered eye or brain damage which makes red things look yellow to them, or which has somehow scrambled whatever neural assemblies are associated with their capacity to use colour terms. Of course, this supposition is far-fetched; but it appears at least to make sense, and that is all that is required. Reformulating the argument once again, Wright claims:
386
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
none of us, if he finds himself on his own about a new candidate for <(>-ness, and with no apparent way of bringing the rest of us around, can sensibly claim to recognize that the community has here broken faith with its antecedent pattern of application for <$>; the proper conclusion for him is rather that he has just discovered that he does not know what <|> means.44
The italicized words are evidently crucial, since Wright's claim would clearly be preposterous without them. Even with them, the claim that a charge of community-wide error simply makes no sense would be unwarranted if the key words meant merely that there does not appear to be any way in which the isolated individual can persuade the rest that they have gone wrong. Earlier, Wright speaks of the individual being incorrigibly out of line, suggesting that there is, without qualification, no way in which he can bring the community around. But now, it seems to me, we need to be a lot clearer about just what supposition it is that we are being invited to entertain, before we can say what follows. Why is it that he can't do that? Are we also to suppose that the situation is, as it were, symmetrical - so that there is, equally, no way in which the community can bring the individual round to its way of thinking? If so, then unless it is further being assumed that one or other party is the victim of some cognitive malfunction which, however, mysteriously resists exposure, we are in effect being asked to suppose that the individual goes one way and the community the other, without either being cognitively at fault: but then it seems that the supposition effectively begs the key question, of whether or not there is a fact of the matter to be recognized.
Wright agrees that his earlier argument needs reinforcement at this point, and seeks to provide it in the later, refurbished version. This proceeds in two stages, corresponding to a distinction Wright draws between basic statements and others. Very roughly,45 the former are statements involving only demonstratives together with concepts whose mastery consists in the possession of some appropriate recog-nitional capacity - concepts for which "competent use standardly presupposes no more than normal sensory capacities and ostensive teaching", such as concepts of colour, taste or pitch. In the first stage, Wright deploys a strengthened version of the argument we have been considering, restricted in scope to basic statements; the second stage then generalizes the conclusion - if basic statements lack objectivity of meaning, so must the remainder. The first-stage argument proceeds, in essentials, as before. But now Wright adds a supplementary argument to close off the gap opened up by the apparent possibility that the lone dissenter is right, the rest of his community having indeed gone astray in their application of basic concepts, perhaps as a result of the deleterious effects of some environmental contaminant upon their capacity to apply them reliably, or for some similar reason. What, in more detail, would it be like, he asks, for there to be available reason to think that everyone (else) had gone astray in their application of basic concepts?
Wright's argument starts from the idea that, if this is a genuine possibility, we should be able to see how a sustainable case could be made for thinking it to have been realized. We may suppose that the lone dissenter can put up a case of this sort: he points to (a) evidence that the rest have been exposed to a certain environmental
387
BOB   HALE
contaminant and (b) evidence that when others have been exposed to this contaminant in the past, their basic judgements of the relevant sort have been distorted. Against such a case, a doubt of the following kind may be raised: the evidence (b) involves the claim that the affected subjects' basic judgements were distorted, and the basis for this claim is that those judgements were found to be at odds with basic judgements made by others who were not affected. The case assumes that we are warranted in taking it that the judgements made by those who were not affected were indeed correct. But might not those very judgements themselves have been the product of widespread error? Unless and until adequate reason can be provided to discount this possibility, the case the lone dissenter has sought to make is worth nothing. Wright's counter-claim is, in effect, that the possibility could only rationally be discounted by appeal to something like this principle, as being analytic of the notion of basic statement:
If there is widespread non-collusive agreement on the truth of a basic statement S and there is adequate reason to suppose that the parties to this widespread agreement understand the concepts involved in S, and are functioning normally in normal conditions for exercise of the appropriate recognitional capacities, and there is no further evidence germane to the case, then anyone apprised of all these facts has adequate grounds for regarding S as true.4f>
The snag is that the objectivist about meaning can hardly regard this principle as analytic; on her view, it can be at best a contingent truth, and that will not be enough to see off the challenge. In short, the attempt to sustain the contractual model by appealing to the possibility of widespread communal error opens the doors to scepticism.
I shall not here try to evaluate this argument; Wright has, in my view, made a powerful case which - so far as I know - has yet to receive an effective reply. And if he is right, the case against objectivity of meaning relies, in its final form, on no specifically anti-realist premiss. I leave the reader to ponder it, and turn instead to my second main - and by now, pressing - question.
4.5 Investigation-independence and objectivity of judgement
The term 'investigation-independent fact' is indeed strongly suggestive of a familiar enough and, for all that it calls for philosophical articulation, seemingly indispensable notion of objectivity; so much so that the suggestion that we should deny that there are any such facts may strike us as the philosophical equivalent of red-rag-waving. I shall try to explain why the bulls should stand their ground.
Preparatory to introducing the notion of objectivity of judgement - as distinct from that of objectivity of meaning - Wright says:
Cognition is relational: it is a matter of arriving at true opinions in a manner sensitive to states of affairs whose obtaining is somehow independent of one's so arriving. Moreover, such a sensitivity must be conceived as essentially fallible.47
388
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
Obviously the crucial words here are 'somehow independent'. In what does the independence of cognized states of affairs consist? Part of what is involved, at least, is that in any particular case where a subject S comes to know (and so forms a true opinion) that p, it should be the case that the state of affairs in virtue of which it is true that p does not depend in any way at all on S's coming to believe that p; or, indeed, on S's or anyone else's coming to hold any opinion on the matters in question. That is, we conceive of the relevant state of affairs as such that its obtaining is consistent with universal ignorance of its doing so. That is one component in the notion of objectivity of judgement, as Wright characterizes it. This is naturally expressed in counterfactual terms: whenever a subject S is properly described as coming to know that p, it would (still) have been the case that p, even if neither S nor anyone else had investigated the matter, or formed any opinion on it. It seems to follow that endorsement of objectivity of judgement for a type of statement entails accepting that there are relevant states of affairs which obtain or not, independently of investigation, in this sense at least: it is the case that p (or not) independently of whether anyone ever did or will carry out an investigation to determine whether or not p (and a fortiori, independently of the result of any such investigation, were one (to be) carried out).
Taking in the other component which Wright includes in the idea of objectivity being charted - that is, the essential fallibility of judgement - the objectivist about judgement is committed to there being true claims of this sort:
(1)    It is the case that p & it would (still) have been the case that p, even if no one had carried out an investigation to determine whether or not p, and even if someone had carried out an investigation, but one that issued in the verdict that not-p.
How about the case where we are concerned with some decidable, but as yet uninvestigated matter? What kind of claim should the objectivist about judgement make then? Well, suppose the question is whether some large integer k is or is not prime. Then the objectivist can say this:
ic is either prime or not. If k is prime, then even if no one ever investigates, it is prime, and were anyone to investigate but come up with a different answer, she would be mistaken; and if ic is not prime, then again, even if no one ever investigates, it is not prime, and were anyone to investigate but come up etc.
Generally, where decidable but as yet uninvestigated matters are in question, the objectivist about judgement may register the sense in which they concern objective states of affairs by asserting an appropriate statement of the form:
(2)    Either p or not-p. If p, then even if no one ever investigates, p, and were anyone to investigate but come up with any other answer, she would be wrong; and if not-p, then again, even if no one ever investigates, not-p, and etc.
389
BOB   HALE
The crucial point for present purposes is that counterfactuals of the kind embedded in these claims are precisely not counterfactuals about what expressions it would be (have been) correct to apply in certain circumstances. As such, they contrast sharply with the kind of counterfactual in terms of which investigation-independence is characterized by Wright - "the [investigation-independent] fact about the object's shape is a fact about how we would describe it if... we continued to use germane expressions in what we regard as the correct way."48
An objectivist about judgement can assert claims of both kinds of the forms (1) and (2). This marks one clear sense in which he can regard certain statements as being true or false in virtue of states of affairs obtaining, or not, independently of investigation. If that is right, then we should question McDowell's right to the following transition, integral to the argument by which he persuades himself that a "familiar and intuitive notion of objectivity" requires the contractual conception of meaning:
The idea at risk is the idea of things being thus and so anyway, whether or not we choose to investigate the matter, and whatever the outcome of any such investigation. That idea requires the conception of how things could correctly be said to be anyway - whatever, if anything, we in fact go on to say about the matter.49
For endorsement of conditionals of types (1) and (2) seems quite enough to hit off the idea of objectivity (of things being thus and so anyway . . .). But those conditionals - at least on the face of it - say nothing about how things could correctly be said to be.
It may be replied that this is a mere artefact of formulation: surely the objectivist ought not to make any bones about accepting these reformulations:
(3)    It is true to say that p & it would (still) have been true to say that p, even if no one had carried out an investigation to determine whether or not p, and even if someone had carried out an investigation, but one that issued in the verdict that not-p.
(4)    Either it is true to say that p or it is true to say that not-p. If it is true to say that p, then even if no one ever investigates, it is true to say that p; and if it is true to say that not-p, then again, even if no one ever investigates, it is true to say that not-p.
Well, of course he should accept them, since their acceptability is guaranteed by the equivalence of'p' with 'it is true to say that p'. But the effect of securing McDowell's first transition by appeal to the equivalence thesis is simply to put in question the next transition in his argument; that is, from the second sentence, just quoted, to:
and this notion of correctness can only be the notion of how a pattern of application that we grasp . . . extends, independently of any investigation, to the relevant case.
390
ROLE-FOLLOWING,   OBJECTIVITY   AND   MEANING
This transition would be good if, but only if, we could pass from (3) to:
(5) It is true to say that p & it would (still) have been correct to assert 'p\ even if no one had carried out an investigation to determine whether or not p, and even if someone had carried out an investigation, but one that issued in the verdict that not-p.
or something to that effect. For it is only if some such transition to a (counterfac-tual) claim about what words it would have been correct to use is allowable that endorsement of objectivity in the sense of the premiss can be made out to involve commitment to the idea of a pattern of application (of some words) extending independently of investigation. But it should be quite clear that the object-linguistic counterfactual simply does not entail its metalinguistic counterpart. In short, McDowell's argument is vitiated by a simple equivocation on 'how things could correctly be said to be'. The second step in his argument is good only if this says something about what words it would be correct to use; the first is good only if it does not.
If what I have said is right, there is after all a gap, discernible by one who rejects objectivity of meaning, between the truth of a statement and its actual verification. Contrary to first appearances, denying that there are investigation-independent facts (in the sense in which Wright does deny this) does not involve denying that, when we correctly perform an elementary calculation, the correctness of our result is independent of our performance. In that sense we can agree with Dummett that there is, in advance of our carrying it out, a determinately correct result. And more generally, when we make a valid inference from true premisses to the conclusion that p, the truth of our conclusion does not wait upon our coming to it; it can, without presupposing the contractual model, be acknowledged that it would still have been the case that p, even if we had not drawn the inference. In that sense, by making the inference we acquire knowledge of a fact of which we had previously been ignorant but which was already there to be known. There is, to be sure, more to be said before we can lay claim to a satisfying explanation of the usefulness of deductive inference.50 But this much is, it seems to me, enough to dispel the appearance that no such account can be forthcoming if the contractual model of meaning is abandoned.
5 Concluding remarks
I have concentrated here on two discussions, both of which enlist Wittgenstein's rule-following considerations in support of radical and highly revisionary conclusions about the objectivity of meaning - conclusions which may appear to entail, and have been taken to entail, consequences for the objectivity of truth and judgement which are no less radical and revisionary. My principal concern has been to argue that, however unpalatable these conclusions - Kripke's semantic irrealism and Wright's rejection of the contractual model and investigation-independence -may seem, we have as yet no compelling demonstration of their unacceptability,
391
BOB   HALE
and thus have no advance right to think that the arguments leading to them must be unsound. I should like to conclude with some remarks about how we should view the situation.
First, whilst we have as yet no decisive ground for thinking semantic irrealism unstable, it is quite another question whether any argument Kripke gives, or might have given, compels its acceptance. The greater part of Kripke's argumentation -and certainly the most convincing part of it - is directed against attempts to explain, in naturalistically reductive terms, what it is to mean something by an expression. To the extent that it is effective, it secures its effect by taking undisputed features of the concept of meaning - generality of application and normativity - and showing that they elude explanation on the proposed naturalistic basis. Clearly no argument of this kind could undermine a view according to which semantic, or more generally intentional phenomena are irreducible. To establish semantic irrealism requires no less than a demonstration that indispensable features of the concept of meaning cannot be jointly instantiated. The closest Kripke comes to providing one is in his dismissive discussion of the idea that meaning something is a sui generis 'unique introspectible state'; but this comes nowhere near to a demonstration that severally essential ingredients in the concept of meaning cannot coexist. Furthermore, the required features are, as Wright reminds us, apparently coherently co-exemplified in our standard intuitive notion of intention. We have no a priori guarantee that that notion could not turn out to be incoherent; but no reason to think it so is yet in sight.
Matters stand otherwise, it seems to me, with Wright's conclusion. Here we are faced with a prima facie compelling argument for the bankruptcy of the contractual model to which, so far as I have been able to see, no effective counter has been provided. And if I am right, horrified reactions to the ensuing rejection of investigation-independence can be seen to be misplaced, once that notion is properly separated from a more modest notion of objectivity which can perfectly well survive without contractual underpinning. What does then become pressing is the need for a satisfying, detailed account of how we may view meaning as - in Wright's own, somewhat opaque phrase - 'shaped by features of our ongoing linguistic behaviour'. This is one direction in which we have a good way yet to travel, before we can reckon ourselves to have appreciated the full significance of Wittgenstein's remarks on rule-following.
Notes
1    Wittgenstein (1967. §§ 138-242; 1978 Part VI).
2    McDowell (1984), Baker and Hacker (1984a and b; 1985).
3    Kripke (1982), Wright (1980; 1981; 1984b), Carruthers (1984).
4    See, in addition to the works cited in the two preceding notes, Peacocke (1981), McGinn (1984), Budd (1984), Pears (1988, chs 16-18), Williams (1991), Luntley (1991) and Wright (1989b. pp. 239-45).
Another important focus of controversy which must be left unexplored here is the exact relationship between Wittgenstein's discussion of rule-following and his arguments against the possibility of a private language. See Chapter 6, meaning and privacy.
392
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
5    See Kripke (1982, pp. 14-15), Wright (1984a, pp. 762-3) and Boghossian (1989, p. 515).
6    This calls for some qualification: whilst there is, so far as I know, no published attempt to show that perverse Kripkean hypotheses break down when we try to work them through in detail, at least one critic - Neil Tennant - attempts to make such a case in an as yet unpublished sequel to his Anti-Realism and Logic.
7    Kripke (1982, pp. 26-28).
8    Ibid., pp. 28-32.
9    Ibid., p. 55.
10    Cf. Wittgenstein (1967, §258).
11    Kripke (1982, p. 14).
12    For dispositional theories see, for example, Papineau (1987), Fodor (1987; 1991). For doubts about the efficacy of Kripke's objections, see Blackburn (1984a) and Forbes (1983). Attempts to uphold a causal account of meaning or reference in the face of the sceptical argument may be found in Goldfarb (1985) and McGinn (1984, pp. 164-6), who also defends the capacity proposal (pp. 168-75). Chomsky (1986) argues that Kripke improperly restricts the search for meaning-consititutive facts - the claim that you formerly followed a certain rule is a theoretical claim, and as such answerable to future evidence, as well as evidence concerning, for example, your past linguistic behaviour or conscious mental life. Boghossian (1989) provides a useful survey and assessment of attempts at a straight solution which accept Kripke's reductionist assumption. See Chapter 5, a guide to naturalizing semantics.
13    For criticism of Kripke's reductionist leanings and proposals for non-reductive solutions, see Boghossian (1989, pp. 540-9), McGinn (1984, pp. 150-64) and Wright (1984a, pp. 772-7).
14    Kripke (1982, p. 60).
15    Ibid., p. 86. See Chapter 13, theories of truth, sect. 8.
16    As some commentators, e.g. Blackburn (1984a, p. 285), have remarked.
17    I cannot, for reasons of space, pursue the matter here - for illuminating discussion, see Wright (1992) passim. See also Chapter 12, realism and its oppositions, final section.
18    Cf. Wright (1984b) and Boghossian (1989; 1990).
19    McDowell (1984, p. 330).
20    Cf. Wright (1984a) and Boghossian (1989; 1990).
21    Wright (1984a, p. 769).
22    Ibid., pp. 769-70.
23    Short version, Boghossian (1989); full dress, Boghossian (1990).
24    Boghossian (1989, p. 526).
25    Boghossian (1990, p. 165, n. 17).
26    Cf. Wright (1992a, p. 234; 1993, pp. 318-24).
2 7 The difficulty has been stated with respect to Wright's version of the globalizing argument - but, as Wright himself points out, somewhat similar troubles afflict Boghos-sian's version: cf. Wright (1992a, pp. 218-20).
28    Blackburn (1989) makes a similar objection to Boghossian's version of the argument. Doubts about the sufficiency of this kind of response to the globalizing argument are developed in Wright (1992a, pp. 222-6).
29    In his 1980, 1981 and 1984b.
30    In the later paper Wright refers to this doctrine, according to which meanings are conceived of as "fully settled by over-and-done-with behavioural and intellectual episodes", as that of objectivity of meaning.
393
BOB   HALE
31    Cf. Wright (1981, p. 99).
32    Wright (1981, p. 100).
3 3 Wright conducts this stage of the argument in terms of the question whether sense can be given to the claim that an individual is being faithful to an idiolectic pattern of application. But as I understand it, what really matters, for this part of the argument, is not whether she is viewed as seeking to suit her performance to a pattern peculiar to herself rather than a shared pattern, but whether others are supposed to be able to evaluate her performance.
34    In Wright (1984b).
35    Cf. Wright (1980, pp. 219-20).
36    At best, as Wright observes (1980, p. 219), it is far from clear how the probability of communal agreement being in or out of line with the requirements of the ratification-independent pattern could be assessed.
37    (1980, p. 219), my emphasis.
38    Cf. Wright:
Of course, it may happen that the community changes its mind; and when it does so, it does not revise the judgement that the former view enjoyed consensus. But that is a fact about our procedure; to call attention to it is to call attention to the circumstance that we make use of the notion that we can all be wrong, but it is not call attention to anything which gives sense to the idea that the wrong-ness consists in departure from a ratification-independent pattern. (1980, pp. 219-20)
39    Wright (1980, p. 221).
40    McDowell (1984, p. 325). Whether this is a fair assessment of the import of Wright's argument, and whether, in particular, McDowell is right to identify the familiar intuitive notion of objectivity to which he gives expression with that of investigation-independence, are questions to which we must shortly return.
41    The scattered quotations from McDowell are all taken from his 1984, p. 336.
42    In effect, because Dummett is not explicitly discussing Wright's argument, though it may be that he does in fact have that argument in mind.
43    The quotations are from Dummett (1994, pp. 63-4).
44    Wright (1980, p. 218).
45    This is a very rough description. For a much more careful account, see Wright (1984b, pp. 276-83). A needed further refinement of the notion of basic statement is provided in Wright (1992b, pp. 40-2).
46    Cf. Wright (1984b, p. 288).
47    Ibid., p. 281.
48    Wright (1980, p. 216).
49    McDowell (1984, p. 325).
50    For further discussion, see Dummett (1973; 1991a, pp. 36-42 and 305-6 and 1991b, ch. 7).
51    Thanks to Jim Edwards and Crispin Wright for very helpful comments.
References and suggested reading
Baker, G. and Hacker, P.M.S. 1984a: On misunderstanding Wittgenstein: Kripke's private
language argument. Synthese, 58 (3), 407-50. ------1984b: Scepticism, Rules and Language. Oxford: Blackwell.
394
RULE-FOLLOWING,   OBJECTIVITY   AND   MEANING
-----1985: Wittgenstein: Rules, Grammar and Necessity. Oxford: Blackwell.
Blackburn, S. 1984a: The individual strikes back. Synthese, 58 (3), 281-301.
-----1984b: Spreading the Word. Oxford: Clarendon Press.
-----1989: Wittgenstein's irrealism. In Rudolf Halier and Johannes Brandl (eds), Wittgenstein - Towards a Re-Evaluation. Vienna: Hölder-Pichler-Tempsky, pp. 13-26. Boghossian, P.A. 1989: The rule-following considerations. Mind, 93 (392), 507-49.
-----1990: The status of content. Philosophical Review, 99 (2), 157-83.
Budd, M. 1984: Wittgenstein on meaning, interpretation and rules. Synthese, 58 (3), 303-
23. Carruthers, P. 1984: Baker and Hacker's Wittgenstein. Synthese, 58 (3), 451-79. Chomsky, N. 1986: Knowledge of Language. New York: Prager. Dummett, M. 1973: The justification of deduction. In Dummett's Truth and other enigmas.
London: Duckworth 1978.
-----1991a: Frege: Philosophy of Mathematics. London: Duckworth.
-----1991b: The Logical Basis of Metaphysics. London: Duckworth.
-----1994: Wittgenstein on necessity: some reflections. In Peter Clark and Bob Hale (eds),
Reading Putnam. Cambridge, Mass. and Oxford: Blackwell. Fodor, J. 1987: Psychosemantics: the problem of meaning in the philosophy of mind. Cambridge,
Mass. and London: MIT Press. -----1991: A theory of content. In A Theory of Content and Other Essays. Cambridge, Mass.:
MIT Press. Forbes, G. 1983: Scepticism and semantic knowledge. Proceedings of the Aristotelian Society,
84,223-37 Fogelin, R. 1976: Wittgenstein. London: Routledge and Kegan Paul. Goldfarb, W. 1985: Kripke on Wittgenstein on rules. Journal of Philosophy, 82, 471-88. Holtzman, S.H. and Leich, CM. (eds) 1981: Wittgenstein: To Follow a Rule. London:
Routledge and Kegan Paul. Kripke, S.A. 1982: Wittgenstein on Rules and Private Language. Oxford and New York:
Blackwell. Luntley, Michael 1991: The transcendental grounds of meaning and the place of silence. In
Puhl (1991), 170-88. McDowell, John 1984: Wittgenstein on following a rule. Synthese, 58 (3), 325-63. McGinn, Colin 1984: Wittgenstein on Meaning. Oxford: Blackwell. Papineau, David 1987: Reality and Representation. Oxford: Blackwell. Peacocke, Christopher 1981: Rule-following: the nature of Wittgenstein's arguments. In
Holtzman and Leich (1981). Pears, David 1988: The False Prison. Vol. 2, Oxford: Oxford Unversity Press. Puhl, Klaus (ed.) 1991: Meaning scepticism. Berlin: de Gruyter. Putnam, Hilary 1981: Convention: a theme in philosophy. New Literary History, 13 (1),
reprinted in his Realism and Reason: Philosophical Papers. Vol. 3, Cambridge: Cambridge
University Press (1983). Williams, Meredith 1991: Blind obedience: rules, community and the individual. In Puhl
(1991), 93-125. Wittgenstein, Ludwig 1967: Philosophical Investigations. 3rd edn, Oxford: Blackwell.
------1978: Remarks on the Foundations of Mathematics. 3rd edn, Oxford: Blackwell.
Wright. Crispin 1980: Wittgenstein on the Foundations of Mathematics. London: Duckworth. ------1981: Rule-following: objectivity and the theory of meaning. In Holtzman and Leich
(1981).
395
BOB   HALE
Wright, Crispin 1984a: Kripke's account of the argument against private language. Journal of Philosophy, 81 (12), 759-78.
------1984b: Rule-following, meaning and constructivism. In Charles Travis (ed.), Meaning
and Interpretation. Oxford: Blackwell 1986, pp. 271-97.
------1989a: Review of McGinn 1984. Mind, 97 (390), 289-306.
------1989b: Wittgenstein's rule-following considerations and the central project of theoretical linguistics. In Alexander George (ed.), Reflections on Chomsky. Oxford Blackwell, pp. 233-64.
------1992a: Truth and Objectivity. Cambridge, Mass. and London: Harvard University Press.
------1992b: Scientific realism and observation statements. In David Bell (ed.), Science and
Subjectivity. Berlin: Academie Verlag, pp. 21-46.
------1993: Eliminative materialism: going concern or passing fancy? Mind and Language, 8
(2), pp. 316-26.
396
Blackwell
Companions to
Philosophy
A Companion to
the Philosophy of
Language
Edited by
BOB HALE
and
CRISPIN WRIGHT
BLACKWELL